OpenAI Gives ChatGPT A New ‘Memory’ Feature
From a Systems Engineering Perspective
OpenAI is an artificial intelligence company that created a chatbox named Chat Generative Pre-trained Transformer, or ChatGPT. It was launched on November 30, 2022, to enable users to direct a conversation towards a desired format, style, level of detail, and desired length. ChatGPT uses natural language processing to create human-like conversational dialogue. The model can respond to various questions and formulate content such as articles, essays, codes, and social media posts. This new technological advancement quickly became extremely useful, and in July of 2023, OpenAI gave ChatGPT a new memory feature.
The company has released long-term memory in ChatGPT, which is a “function that maintains a memory of who you are, how you work, and what you like to chat about”(Goode, 2024). This feature has been named ‘Memory’ and allows the user to have ChatGPT remember specific details or details. This feature can be turned off at any time, and the user also can have ChatGPT delete any information it has stored. There is also a feature called ‘Temporary Chats’ that allows users to have certain conversations without using the memory feature. Overall, this new feature in ChatGPT allows the model to retain context and information from previous conversations, enabling it to generate responses that are more coherent and contextually relevant.
The stakeholders for a new feature like this include the users, developers and engineers, product managers, and even data privacy experts. Users are the individuals or organizations that will be directly using the feature. In the case of ChatGPT’s memory feature, users might be individuals looking to store information or context for later retrieval during their conversations. The developers and engineers team are responsible for building and maintaining the feature. This includes software developers and machine learning engineers who work on implementing and refining the memory feature. Product managers are responsible for overseeing the development of new features, gathering requirements, and ensuring that the feature meets the needs of users and aligns with the overall product strategy. Lastly, given that the memory feature involves storing and retrieving user data, stakeholders with expertise in data privacy and security would be crucial to ensure that user data is handled responsibly and under relevant regulations.
The ChatGPT memory feature can be understood and analyzed through various systems engineering perspectives. From a requirements perspective, the INCOSE SE handbook emphasizes the importance of managing requirements throughout the system’s life cycle. The new memory feature would be motivated by the need and ability to improve the conversational abilities of the model. The development process would be responsible for any requirements regarding context retention, recall accuracy, memory capacity, and response coherence. Requirements could also be related to storage capacity or other memory-related criteria that the system must meet. From a system architecture perspective, the new memory feature could be a part of these design considerations. Determining the type of memory needed, the location within the system, how memory is shared among components or any interfaces for accessing memory would all be part of the system architecture design considerations. The new memory feature impacts the architecture of the system by requiring the inclusion of memory mechanisms within the pre-existing model. This would include designing new data structures to store and retrieve contextual information.
Some other system engineering perspectives would be lifecycle considerations and verification and validation. From a lifecycle standpoint, this memory feature would extend into ongoing maintenance or updates. This would include monitoring the performance of this memory feature and addressing any limitations or issues that could occur. This feature allows for continuous improvement based on user feedback and any evolving requirements. The INCOSE SE handbook promotes a holistic view of systems engineering, and this memory feature would be considered throughout the system’s entire operation or maintenance. From a verification and validation perspective, ensuring that the memory feature meets specific requirements and operates as intended would be the focus. This would include evaluating, validating, and testing the model. The INCOSE SE handbook emphasizes on the importance of verifying and validating the system’s requirements. This memory feature can have many subsystems that need verification and testing to ensure they meet specified requirements.
Overall, OpenAI’s new memory feature on ChatGPT has been utilized to provide services for the benefit of users and other stakeholders. The process and evolution of this feature have been influenced by system engineering. The INCOSE SE handbook provides a wide range of system engineering practices. A variety of these practices were essential for OpenAI’s development of this new memory feature on ChatGPT.
References:
Goode, L. (2024, February 13). OpenAI gives chatgpt a memory. Wired. https://www.wired.com/story/chatgpt-memory-openai/
Memory and new controls for CHATGPT. Memory and new controls for ChatGPT. (n.d.). https://openai.com/blog/memory-and-new-controls-for-chatgpt
Metz, C. (2024, February 13). OpenAI gives chatgpt a better “memory.” The New York Times. https://www.nytimes.com/2024/02/13/technology/openai-gives-chatgpt-a-better-memory.html
Walden, D. D. (2015). INCOSE Systems Engineering Handbook a guide for systemlife cycle processes and activities. Wiley.
YouTube. (2024a, January 26). GPT-4’s new “memory” feature is stunning (CHATGPT memory). YouTube. https://www.youtube.com/watch?v=-U4o7ZnLpPw
YouTube. (2024b, February 13). GPT-4’s new “memory” feature is released! (Chatgpt memory update). YouTube. https://www.youtube.com/watch?v=EaNyFB-jctg
Picture URLs:
https://redis.com/blog/chatgpt-memory-project/
